Pumping sets
Seals that start dripping and couplings that go out of alignment.
Teal Neural
The machine warns you before it breaks.
Signs that can be seen but that nobody is looking at at three in the morning.
The fault starts out as a detail.
Before a machine stops altogether there are usually weeks of small signs: a vibration that was not there, a tiny leak, a part that no longer returns exactly to its place.
All of them can be seen. None of them sets anything off, because alarm systems are built for values out of range, not for things that change little by little.
And preventive maintenance runs on the calendar, not on condition: the part is changed when it is due, whatever state it is in.
Comparing today with last month.
The camera is fixed, so today's image and the one from weeks ago are the same field of view. That lets you see what the eye misses from one day to the next.
You mark what has to be watched — a seal, a travel path, a rest position — and the system alerts you when it drifts from how it was.
The technician's round still happens. What changes is that it arrives knowing where to look and with the comparison in front of it.
What it watches.
- Leaks and drips that start small and grow.
- Parts that do not go back to their rest position.
- Abnormal vibration, seen as movement in the image.
- Dirt, swarf or waste building up where it should not.
- Visible wear: belts, chains, guides.
- Component temperature, with a thermal camera.
- Confirmation that the work was done: guard closed, grease applied.
- Time since the last piece of work on each machine.
Lo que Teal Neural entiende
5 detecciones grabadas.
Producción y línea de fabricación.
Teal
Teal
Teal
Teal
Teal
Cada una es una grabación del sistema funcionando. Pasa el ratón por encima —o tabula hasta ella— para verla en movimiento.
A real case.
A pumping set had a seal that started dripping in March. The April round did not note it — it was a small stain — and neither did the May one.
The image comparison flagged it at three weeks: the stain had gone from two centimetres to twelve. The seal was changed during a planned shutdown, instead of during the unplanned one in June.
Where it is used.
Drives
Belts and chains with visible wear, and pulleys off their travel path.
Plant rooms
Unattended equipment that nobody enters between rounds.
Remote sites
Places where going out to look costs half a day.
Checking that it was done.
A signed maintenance report says the work was carried out. It does not say the guard was left closed, that the grease went in, or that the tool was not left inside.
Those are things you can see. Comparing the image before and after a job checks the final state, and it is recorded with the date.
It is not mistrust: it is that at three in the morning, after four hours, anyone leaves a guard open.
From preventive to condition-based.
Preventive maintenance runs on the calendar: the part is changed every six months, whatever state it is in. That means changing good parts and, now and then, arriving late to a bad one.
With a series of dated images of the real condition, the decision stops being the calendar and becomes the curve: this belt has looked the same for three months and this one has frayed in a fortnight.
It does not replace the maintenance plan. It gives it a figure it does not have today.
What it takes to make it work.
- A fixed camera. If it moves, you lose the comparison with last week.
- The item to be watched visible, and at a size where the change shows.
- Steady lighting, or at least lighting that is comparable at the same time of day.
- For temperature, a thermal camera: a normal camera does not measure degrees.
Casos grabados
Funcionando, en sitios de verdad.
Botellas atascadas - Optimiza tu línea de producción con Infinity Neural
Optimización de la Producción Frutícola con Teal Neural
Monitorización de manómetros con Teal Neural
Teal Neural: Optimizando la producción industrial
Teal Neural: Optimización y control en cintas de transporte
What it does not do.
It does not really measure vibration: it sees that something is oscillating more than before, not how many hertz. If the figure has to be exact, that is an accelerometer.
And it needs a fixed camera. If somebody moves it, or it pans on its own, the comparison with last week is lost and you have to start again.
And what sits alongside it.
Other things the same family understands, on the same cameras.
What people ask us.
▸Does it replace predictive maintenance?
No. Real predictive maintenance is done with vibration, temperature and oil analysis, and those are sensors.
This covers what can be seen and what nobody looks at between rounds today: a leak starting, a part that has shifted, dirt building up where it should not.
▸How often does it compare?
As often as you tell it to. For things that change over weeks, a daily comparison is enough and avoids noise. For what can change in minutes — a spill — the alert is immediate.
▸What happens if somebody moves the camera?
The reference is lost and it has to be set again. The system detects it — the whole image changes at once — and tells you it has happened, instead of treating a comparison that no longer holds as valid.
▸Will it work with the cameras we already have?
If they are fixed and the item to be watched is seen at a size where the change shows, yes. We look at it with an image before saying anything.
Get started
Do you want this on your site?
This is already recorded working. What is left is to see whether your camera allows it: where it is fitted, what angle it has and what can be seen from there. If it will not work, we will tell you.